Extracellular ATP increases cation fluxes in human erythrocytes by activation of the P2X7 receptor.

Canine erythrocytes are known to undergo a reversible increase in cation permeability when incubated with extracellular ATP. We have examined the expression and function of P2X receptors on human erythrocytes using confocal microscopy and a panel of anti-P2X(1-7) antibodies and have measured monovalent cation fluxes in the presence of various nucleotide agonists. Human erythrocytes expressed P2X7 receptors on all cells examined from eight of eight subjects, as well as P2X2 at a far lower staining intensity in six of eight subjects. ATP stimulated the efflux of 86Rb+ (K+) from human erythrocytes in a dose-dependent fashion with an EC50 of approximately 95 microM. Other nucleotides also induced an efflux of 86Rb+ from erythrocytes with an order of agonist potency of 2'- and 3'-O(4-benzoylbenzoyl) ATP (BzATP) > ATP > 2-methylthio-ATP (2MeSATP) > adenosine 5'-O-(3-thiotriphosphate) (ATPgammaS), whereas ADP or UTP had no effect. ATP-induced efflux of 86Rb+ from erythrocytes was inhibited by extracellular Na+ and oxidized ATP, as well as by KN-62, an antagonist specific for the human P2X7 receptor. When erythrocytes were incubated in isotonic KCl medium, the addition of ATP stimulated an 86Rb+ influx approximately equal in magnitude to ATP-stimulated 86Rb+ efflux from the same cells. BzATP also stimulated the influx of 22Na+ into erythrocytes incubated in isotonic NaCl medium. Both ATP-induced efflux and influx of 86Rb+ and 22Na+ were impaired in erythrocytes from subjects who had inherited loss-of-function polymorphisms in the P2X7 receptor. These results suggest that the reversible permeabilization of erythrocytes by extracellular ATP is mediated by the P2X7 receptor.

Many of our concepts of cellular Na ϩ and K ϩ homeostasis were based on experiments in the erythrocyte, a cell type in which intracellular Na ϩ and K ϩ concentrations could be readily changed and the subsequent effect on ion transport could be measured. Early studies of the sodium pump used the technique of hypotonic hemolysis and resealing of ghosts to study the dependence of Na ϩ pumping on intracellular cation concentration (1,2), whereas the first description of Na ϩ -K ϩ -2Cl Ϫ cotransport owed much to a nystatin technique to alter the Na ϩ gradient and change the driving force for cotransport (3). Yet another technique was introduced by Parker et al. (4,5) who showed that intracellular Na ϩ and K ϩ concentration of canine erythrocytes could be equilibrated with the cation concentration in the medium simply by incubation with extracellular ATP for 30 -60 min and that removal of ATP restored the basal permeability of the cell. This reversible effect of ATP suggested the involvement of an ATP-gated cation channel, but it is only with recent observations and knowledge of the P2X receptor family (6) that it is possible to study this question.
Seven subtypes of the P2X receptor family have been identified based on a common structure of two transmembrane domains with intracellular amino and carboxyl termini and a trimeric structure in the plasma membrane (6). Although P2X 1 -P2X 6 channels all show desensitization in the continued presence of agonist (6), the converse occurs with the P2X 7 channel, which undergoes slow dilatation over 10 -30 s to a second and larger permeability state allowing a massive loss of cellular K ϩ and a Na ϩ gain (7)(8)(9). A range of downstream events follow P2X 7 activation such as membrane blebbing (10) and microvesicle shedding (11) with the former requiring activation of Rho and p38 mitogen-activated protein kinases and an intact cytoskeleton (12)(13)(14). P2X 7 activation also leads to externalization of phosphatidylserine (11,15) as well as the stimulation of the caspase cascade and a stress-activated protein kinase-dependent pathway resulting in apoptosis (16,17). P2X 7 is variably expressed on the surface of leukocytes being absent on neutrophils, with moderate expression on lymphocytes and the highest expression on monocytes, dendritic cells, and macrophages (18 -20).
We have previously described a loss-of-function polymorphism at nucleotide position 1513 of the human P2X7 gene (1513A 3 C) that changes a glutamic acid to alanine at amino acid 496 (E496A) (21) located within an ankyrin-like repeat (22) or a tumor necrosis factor receptor 1-related death domain (23). Subjects who are homozygous for this polymorphism have grossly impaired ATP-induced Ca 2ϩ and ethidium ϩ influx (21), Rb ϩ efflux (24), as well as ATP-induced cell death (21,25,26). The E496A mutation however does not impair ATP-induced currents in either Xenopus oocytes or HEK293 cells transfected with E496A mutant P2X 7 constructs (27), suggesting that this mutation although not affecting the initial opening of the channel somehow prevents its subsequent dilatation to a pore. More recently, we have identified another but less common loss-offunction single nucleotide polymorphism at nucleotide position 946 of the human P2X7 gene (946G 3 A) that changes an arginine to glutamine at amino acid 307 (R307Q) (28). This residue is located within the ATP-binding pocket, which also includes other positively charged residues (Lys 311 and Lys 193 ) that contribute to the binding of the negatively charged phosphate groups of ATP (29). The R307Q polymorphism in heterozygous dosage impairs ATP-induced Ca 2ϩ and ethidium ϩ influx and Rb ϩ efflux, and in combination with the E496A polymorphism, also in heterozygous dosage, causes a complete loss of these fluxes (28). Using immunocytochemical, pharmacological and genetic approaches we have examined whether P2X 7 receptors are present on human erythrocytes. Our results showed that functional P2X 7 receptors are present on human erythrocytes although at lower abundance than on human lymphocytes.
Erythrocytes-The study was approved by the Wentworth Area Health Service Human Ethics Committee (Penrith, Australia). Peripheral blood was collected in heparin-containing vacutainer tubes from 13 normal volunteers of various genotypes previously identified as either wild-type, heterozygous, or homozygous at nucleotide position 1513 (E496A) of the P2X7 gene (21,33). All subjects except one were wildtype at nucleotide positions 946 (Arg 307 ) and 1729 (Ile 568 ) of the P2X7 gene (28,32). One subject was a double heterozygote for both nucleotide 946A and 1513C (28). Blood was centrifuged at 400 ϫ g for 15 min, and the plasma, platelets, leukocytes, and the upper 10% of erythrocytes were removed. The remaining erythrocytes were washed twice in NaCl medium (147.5 mM NaCl, 2.5 mM KCl, 5 mM D-glucose, 0.1% BSA, 20 mM HEPES, pH 7.5) at 1800 ϫ g for 5 min.
Immunofluorescence Staining and Confocal Microscopy-Erythrocytes or whole blood cells were resuspended in NaCl medium at 5 ϫ 10 7 cells/ml, and 20 l was smeared onto glass slides, air dried for 30 min, and fixed with acetone/methanol for 2 min (34). The slides were washed four times over 10 min with phosphate-buffered saline (PBS) and blocked with PBS, 20% normal horse serum, 0.1% BSA for 20 min before incubation for 45 min with anti-P2X antibody diluted in PBS, 0.2% normal horse serum. The slides were washed three times over 30 min with PBS and incubated for 45 min with fluorochrome-conjugated secondary antibody diluted in PBS, 0.2% normal horse serum. The slides were again washed before embedding in glycerol gelatin mounting medium. Cells were visualized with a Leica (Mannheim, Germany) TCS NT UV Laser Confocal Microscope as described (35). 86 Rb ϩ Efflux Measurements-Erythrocytes from wild-type subjects (unless as otherwise indicated) were loaded with 86 Rb ϩ (5 Ci/ml) at a hematocrit of 40% in NaCl medium for 4 h at 37°C. Cells were then washed three times at 4°C with ice-cold NaCl medium and resuspended in either KCl medium (150 mM KCl, 5 mM D-glucose, 0.1% BSA, 20 mM HEPES, pH 7.5) or NaCl medium at a final hematocrit of 5%. 86 Rb ϩ -loaded erythrocyte suspensions (2.5 ml) were incubated in the absence or presence of nucleotide for 60 min at 37°C. At 0-and 60-min time points, 1-ml samples were overlaid on 0.3 ml of phthalate oil mixture and centrifuged at 8000 ϫ g for 30 s. For time-course studies, 86 Rb ϩ -loaded erythrocyte suspensions (10 ml) were incubated in the absence or presence of 1 mM ATP for 4 h, with 1-ml samples collected at 30 min intervals as above. For studies using the P2X 7 antagonist, OxATP (36), 86 Rb ϩ -loaded erythrocytes resuspended in NaCl medium were pre-incubated in the absence or presence of 300 M OxATP for 60 min at 37°C, washed once with NaCl medium, and resuspended in KCl medium before incubation in the absence or presence of ATP for 60 min. For studies using the specific P2X 7 antagonist, KN-62 (37), 86 Rb ϩloaded erythrocytes resuspended in KCl medium were pre-incubated in the presence of 1 M KN-62 or an equal volume of Me 2 SO vehicle for 15 min at 37°C before incubation in the absence or presence of ATP for 60 min. Samples (700 l) from the supernatant above the oil after centrifugation and a second sample (350 l of 86 Rb ϩ -loaded erythrocytes and 350 l of 0.4% saponin) were collected to measure the level of radioactivity using a Wallac (Turku, Finland) 1480 Wizard 3 Automatic Gamma Counter. Hemolysis after a 60-min incubation was less than 1% as determined spectrophotometrically on cell-free supernatants with Drabkin's reagent (38). 86 Rb ϩ and 22 Na ϩ Influx Measurements-Erythrocyte suspensions (2 ml) at a final hematocrit of 5% in KCl or NaCl medium containing 1 mM furosemide and 100 M ouabain (but no BSA) were incubated with 86 Rb ϩ or 22 Na ϩ (2 Ci/ml) in the absence or presence of 1 mM ATP or 0.2 mM BzATP, respectively, for up to 60 min at 37°C. At 15, 30, 45, or 60 min, cells were washed three times with 10 ml of ice-cold isotonic saline (1200 ϫ g for 60 s at 4°C). Erythrocytes were lysed with 1 ml of 10 M NH 4 OH and 100 l of 1% saponin, the level of radioactivity was measured by ␥ counting, and the hemoglobin concentration was determined spectrophotometrically with Drabkin's reagent. Samples at each time point were performed in duplicate.
Statistical Analyses-The differences in 86 Rb ϩ release were compared using either the unpaired Student's t test for single comparisons to control samples (see Figs. 2 and 4) or analysis of variance for multiple comparisons (see Figs. 5 and 6 and Table I) using SPSS 11.5 for Windows (SPSS Inc, Chicago, IL) with p Ͻ 0.05 considered significant.

P2X 7 Receptors Are Present on Human Erythrocytes-We
have previously detected P2X receptors on paraformaldehydefixed human lymphocytes using a panel of polyclonal antibodies and confocal microscopy (30); however, attempts to fix human erythrocytes immobilized onto poly(L-lysine)-coated coverslips or air-dried onto glass slides with paraformaldehyde were unsuccessful because of a large amount of cell loss (results not shown). Therefore, human erythrocytes were air-dried onto glass slides and fixed with acetone/methanol (34) and stained with a panel of polyclonal antibodies against all seven P2X subtypes. Confocal microscopy with two different anti-P2X 7 polyclonal antibodies (30,32) showed P2X 7 to be present at low to moderate levels on all erythrocytes from eight of eight subjects tested (Fig. 1, A and B; results not shown). Of these subjects, five were wild-type ( Fig. 1A), two were heterozygous (results not shown), and one was homozygous (Fig. 1B) for the E496A polymorphism. Labeling was diffuse, although occasional puncta were observed on erythrocytes from some subjects. P2X 2 was also present on erythrocytes; however, the intensity level of staining was much lower than that of P2X 7 (Fig. 1C) and was near to absent on erythrocytes from two of the eight individuals studied (results not shown). P2X 1 , P2X 3 , P2X 4 , P2X 5 , and P2X 6 were not detected, because the labeling using polyclonal antibodies to these subtypes was similar to that of pre-immune sera ( Fig. 1D; results not shown). To determine the relative level of P2X 7 on erythrocytes compared with leukocytes, whole blood from two wild-type subjects was air-dried onto glass slides, fixed with acetone/methanol, and stained with anti-P2X 7 polyclonal antibodies. Confocal microscopy demonstrated two populations of stained cells within the whole blood smears, which morphologically resembled either erythrocytes or leukocytes, with the intensity level of staining much lower on erythrocytes than that of the mixed leukocyte population (Fig. 1, E and F all time points, ATP induced a significant increment in 86 Rb ϩ efflux from erythrocytes compared with cells incubated in the absence of ATP (Fig. 2). The ATP-induced 86 Rb ϩ efflux was linear up to 60 min, therefore this time point was selected for all subsequent efflux studies.
The dose effect of ATP on cation fluxes in human erythrocytes was then studied. Cells loaded with 86 Rb ϩ were incubated at 37°C with varying concentrations of ATP for 60 min. The efflux of 86 Rb ϩ from erythrocytes after a 60-min incubation ranged from 4.0 Ϯ 0.3% in the absence of a nucleotide up to 9.3 Ϯ 0.5% with 2 mM ATP (Fig. 3). Over this range of ATP concentrations an EC 50 value of 95.4 Ϯ 7.1 M was obtained from the dose-response curve (Fig. 3), which is similar to values observed for either recombinant or native P2X 7 (EC 50 ϭ 85 or 86 M, respectively) (8, 39) but greater than that for P2X 2 (EC 50 ϭ 1 M) (40).
The effect of various nucleotide agonists on 86 Rb ϩ effluxes was then studied. Similar to above (Fig. 2), the release of 86 Rb ϩ from erythrocytes after a 60-min incubation in the absence of nucleotide was 3.6 Ϯ 0.4% (Fig. 4). The P2X 7 agonists, 0.2 mM BzATP and 1 mM ATP, induced a significant efflux of 86 Rb ϩ from erythrocytes incubated in KCl medium with 12.0 Ϯ 1.2 and 9.9 Ϯ 0.9% of the total cell-associated 86 Rb ϩ being released at 60 min, respectively (Fig. 4). Other P2X 7 agonists, 1 mM 2MeSATP and 1 mM ATP␥S, which are both partial agonists (39), also induced 86 Rb ϩ efflux (6.8 Ϯ 0.3 and 4.7 Ϯ 0.6%, respectively), although the latter failed to reach significance compared with 86 Rb ϩ efflux in the absence of the nucleotide (p ϭ 0.15). Thus for human erythrocyte P2X 7 , the agonist potencies follow a rank order of BzATP Ͼ ATP Ͼ 2MeSATP Ͼ ATP␥S, which is identical to the order found for recombinant and native P2X 7 (8,39). ADP and UTP, added at either 1 mM (Fig. 4) or 0.1 mM (results not shown) gave a 86 Rb ϩ release similar to control values, thus excluding a role for many of the P2Y receptors some of which are on avian erythrocytes (41).
Na ϩ Impairs ATP-induced 86 Rb ϩ Efflux from Human Erythrocytes-A characteristic of the P2X 7 receptor is its sensitivity to inhibition by extracellular Na ϩ (42,43). Therefore, nucleotide-induced 86 Rb ϩ release from erythrocytes incubated in either KCl or NaCl medium was compared. 86 Rb ϩ efflux in response to either 1 or 0.1 mM ATP was significantly lower in NaCl Erythrocytes (A-D) or whole blood cells (E and F) from subjects either wild-type at 496 (A and C-F) or homozygous for the E496A polymorphism (B) were incubated with rabbit polyclonal antibodies against P2X 7 (A, B, E, and F), P2X 2 (C), or P2X 1 (D) and subsequently with Cy2-conjugated anti-rabbit IgG antibody before examination by confocal microscopy. Arrows (E and F) indicate leukocytes (from two wildtype subjects) with higher levels of P2X 7 than adjacent erythrocytes. The expression of P2X 7 on erythrocytes and whole blood cells using a sheep anti-P2X 7 polyclonal antibody was similar to that observed using a rabbit anti-P2X 7 polyclonal antibody (results not shown). Pre-immune serum was routinely included as a negative control and demonstrated background labeling (results not shown) similar to that of P2X 1 (D) and P2X 3-6 (results not shown). The calibration bar is 5 m.  (Table I). Similarly, the incubation of cells with 1 M KN-62 significantly inhibited both 1 and 0.1 mM ATP-induced 86 Rb ϩ efflux by over 90% (Table I). Neither Ox-ATP nor KN-62 altered the basal release of 86 Rb ϩ from erythrocytes incubated in the absence of ATP (Table I).
Cells with a Loss-of-Function Polymorphism in the P2X 7 Receptor Have Absent ATP-induced Fluxes-We assessed whether BzATP or ATP could stimulate the efflux of 86 Rb ϩ from erythrocytes of subjects who were homozygous for the loss-of-function polymorphism E496A and whose mononuclear leukocytes lack P2X 7 function (21). Incubation of erythrocytes from these polymorphic subjects with either 0.2 mM BzATP or 1 mM ATP produced a release of 86 Rb ϩ of 1.9 Ϯ 0.0 and 1.9 Ϯ 0.1%, respectively, which was similar to that released from either homozygote or wild-type erythrocytes in the absence of nucleotide (2.0 Ϯ 0.2 and 3.3 Ϯ 0.6%, respectively) (Fig. 6). In the same experiments, BzATP and ATP induced a significant release of 86 Rb ϩ from wild-type erythrocytes of 12.8 Ϯ 1.4 and 9.6 Ϯ 0.5%, respectively (Fig. 6). BzATP and ATP also failed to induce 86 Rb ϩ release from homozygote erythrocytes incubated in NaCl medium (results not shown). P2X 7 Activation Increases K ϩ and Na ϩ Influx into Human Erythrocytes-The P2X 7 cation-selective channel/pore shows little or no selectivity between K ϩ and Na ϩ ions (6). We compared the ability of P2X 7 agonists to stimulate 86 Rb ϩ and 22 Na ϩ influx into erythrocytes from wild-type subjects. To reduce the flux of K ϩ and Na ϩ by other transporters, erythrocytes were incubated in the presence of ouabain, a Na ϩ /K ϩ ATPase inhibitor (1), and furosemide, an inhibitor of the Na ϩ -K ϩ -2Cl Ϫ cotransporter (3). ATP (1 mM) induced an influx of K ϩ into wild-type erythrocytes at a rate of ϳ4.0 mole K ϩ /ml of cells/h (Fig. 7A). This rate of influx is comparable with that of ATP-induced K ϩ efflux, which we estimate to be ϳ6.0 mole K ϩ /ml cells/h, assuming a net ATP-induced 86 Rb ϩ efflux of ϳ6.7% over 60 min (Fig. 4) and an intracellular K ϩ concentration of ϳ90 mole/ml of cells. BzATP (0.2 mM) also induced an influx of Na ϩ into wild-type erythrocytes at a similar rate (ϳ5.2 mole Na ϩ /ml of cells/h; Fig. 7B). In contrast, ATP and BzATP failed to induce an influx of K ϩ (ϳ0 mole K ϩ /ml of cells/h) and Na ϩ (ϳ0.2 mole Na ϩ /ml of cells/h), respectively, into erythrocytes from a subject who was a double heterozygote for two loss-of-function polymorphisms in P2X 7 at amino acid positions 307 and 496 (28) (Fig. 7). DISCUSSION Over three decades ago, Parker and Snow (4) demonstrated that extracellular ATP could reversibly increase the permeability of canine erythrocytes to both K ϩ and Na ϩ ions. This effect of ATP was specific to this nucleotide, because neither ADP nor UTP increased cation fluxes and, moreover, the permeability increase could be impaired by the addition of Mg 2ϩ , suggesting a role for the ATP 4Ϫ species. Here we report a similar phenom-  e p Ͻ 0.01 to a corresponding ATP-treated sample in the absence of antagonist.
f p Ͻ 0.05 to a corresponding sample in the absence of ATP. g p Ͻ 0.05 to a corresponding ATP-treated sample in the absence of antagonist. enon for human erythrocytes and show that P2X 7 mediates this effect of extracellular ATP. We developed two different anti-P2X 7 polyclonal antibodies (30,32), and using immunolabeling and confocal microscopy we showed P2X 7 expression on human erythrocytes. Both polyclonal antibodies recognize non-homologous epitopes in the extracellular domain of the P2X 7 receptor (residues 71-86 and 65-81), and both have been previously shown to label P2X 7 -transfected but not mock-transfected HEK293 cells (29,32). These antibodies can also detect native P2X 7 receptors on human lymphocytes (30), and on the basis of fluorescence intensity we estimate that P2X 7 is at least 10-fold less abundant on human erythrocytes than on lymphocytes (Fig. 1). Similarly, measurements of cation permeability of human erythrocytes by standard isotope flux techniques showed that the magnitude of the ATP-stimulated K ϩ and Na ϩ fluxes are far less for human erythrocytes than for lymphocytes. Many features of the erythrocyte fluxes closely resemble those resulting from activation of the P2X 7 receptor either as the native molecule in mononuclear leukocytes or expressed heterologously in HEK293 cells. First, the EC 50 for ATP (ϳ95 M) and the rank order of agonist potency (BzATP Ͼ ATP Ͼ 2MeSATP Ͼ ATP␥S) were typical of recombinant and native P2X 7 (8,39). Second, ATP-induced 86 Rb ϩ efflux was lower from erythrocytes incubated in NaCl medium compared with KCl medium, and it is well documented that Na ϩ inhibits the activation of P2X 7 (42,43). Third, ATP-induced 86 Rb ϩ efflux was inhibited by two known P2X 7 antagonists, OxATP (36) and the more specific inhibitor, . Fourth, ATP and BzATP produced approximately the same increase in the influx of 86 Rb ϩ and 22 Na ϩ , respectively, into erythrocytes, consistent with P2X 7 being a channel/pore showing little or no selectivity between these two monovalent cations (6). Finally, ATP and BzATP-induced fluxes were absent in erythrocytes of three subjects, two being homozygous for the E496A loss-of-function polymorphism (21) and the third being a double heterozygote for the R307Q and E496A loss-of-function polymorphisms (28).
Activation of native P2X 7 by extracellular ATP produces a nearly complete release of 86 Rb ϩ from human lymphocytes within 5 min (42, 44) and from monocytes within 2 min (24), as well as from murine macrophages (45). In contrast, the ATP-induced 86 Rb ϩ efflux from human erythrocytes was much slower with 9.5 Ϯ 0.3% (n ϭ 19) of the cell-associated 86 Rb ϩ being released after a 60-min incubation with 1 mM ATP. We have shown previously that the level of P2X 7 expression on lymphocytes and monocytes correlates with the level of ATPinduced permeabilization to ethidium ϩ (18), suggesting that the slow rate of ATP-induced 86 Rb ϩ release from erythrocytes is because of the low level of P2X 7 expression on these cells. In addition, the observation that only ϳ10% of the cell-associated 86 Rb ϩ is released after a 60-min incubation with ATP and that beyond this time point the release is not linear with time ( Fig.  2) suggests that ATP induces the release of 86 Rb ϩ from a subpopulation of erythrocytes or that the 86 Rb ϩ is released from an intracellular compartment within these cells. It is of interest that ATP-induced cation fluxes in human erythrocytes are much lower than those in their canine counterparts (4,5,46). In canine erythrocytes, 0.5 mM ATP induced a near to total release of 42 K ϩ after a 40-min incubation (4) suggesting a K ϩ efflux at least one order of magnitude greater than shown in Fig. 2 for human erythrocytes. Sha'afi and colleagues (46) also demonstrated that 1 mM ATP could induce a 7-fold increase in Na ϩ influx into canine erythrocytes resuspended in a phos- FIG. 7. P2X 7 agonists induce K ؉ and Na ؉ influx into human erythrocytes. Erythrocytes from subjects either wild-type at 307 and 496 (E, •) or heterozygous for both R307Q and E496A polymorphisms (‚, ) were resuspended in either KCl medium (containing 86 Rb ϩ , 1 mM furosemide and 100 M ouabain but no BSA) (A) and incubated at 37°C in the absence (E, ‚) or presence (• ,) of 1 mM ATP or NaCl medium (containing 22 Na ϩ , 1 mM furosemide and 100 M ouabain but no BSA) (B) and incubated at 37°C in the absence (E, ‚) or presence (•, ) of 0.2 mM BzATP. 86 Rb ϩ and 22 Na ϩ uptake was calculated as the eq of K ϩ or Na ϩ /ml of erythrocytes, respectively. Results are expressed as the mean from duplicate time points. phate-buffered NaCl medium but had little effect on either human or feline erythrocytes. Thus like other mechanisms of cation transport in mammalian erythrocytes (47), those mediated by extracellular ATP also differ between species. Fig. 6 shows that the ATP-induced 86 Rb ϩ efflux was almost absent from erythrocytes of subjects homozygous for the E496A polymorphism. This polymorphism impairs ATP-induced ethidium ϩ influx and 86 Rb ϩ efflux through the dilated pore in leukocytes (21, 24) but does not impair ATP-induced currents that are mediated via the P2X 7 channel in either Xenopus oocytes or HEK293 cells transfected with E496A mutant P2X 7 constructs (27). Our results suggested that the ATPinduced 86 Rb ϩ efflux occurs via the second and larger permeability state (dilated pore) of the P2X 7 receptor and that isotope fluxes of 86 Rb ϩ through the smaller P2X 7 channel are too small to be reliably measured in this cell type. Moreover, our data showed that ATP-stimulated 86 Rb ϩ efflux and influx are approximately equal for erythrocytes suspended in 150 mM KCl, and both are similar to the BzATP-stimulated Na ϩ influx for cells suspended in a similar concentration of NaCl (147.5 mM). Therefore, these results exclude a role for an ATP-gated rectifier channel in the ATP-induced flux of cations in human erythrocytes.
In addition to P2X 7 , we also detected P2X 2 receptors on human erythrocytes using immunocytochemistry and confocal microscopy. However, despite the P2X 2 channel being permeable to Rb ϩ ions (48), three results suggest that P2X 2 receptors contributed little to the ATP-induced 86 Rb ϩ efflux. First, P2X 2 expression on erythrocytes was lower than P2X 7 , possibly by as much as one order of magnitude, although we can not rule out that the difference in expression between these two purinoceptors may be because of differences in the immunoreactivity of the antibodies used. Second, erythrocytes that lacked P2X 2 , as determined by confocal microscopy, displayed ATP-induced 86 Rb ϩ efflux. Finally, as discussed above, the rank order of agonist potency (BzATP Ͼ ATP Ͼ 2MeSATP Ͼ ATP␥S) and the EC 50 for ATP (95.4 Ϯ 7.1 M) are typical of P2X 7 , whereas for human P2X 2 , the four nucleotides BzATP, ATP, 2MeSATP and ATP␥S are equipotent with an EC 50 of ϳ1 M (40). The failure to detect any functional P2X 2 may simply reflect the low level of expression of this receptor on human erythrocytes. In one study, extracellular ATP stimulated a regulatory volume decrease, which reversed the hypotonic swelling of Necturus erythrocytes via activation of a receptor characteristic of P2X 2 (49,50); however, a similar mechanism could not be observed in human erythrocytes (49). Nonetheless, differences in P2X expression between erythrocytes of different species are likely to be found.
It is likely that P2X 7 has functional significance for human erythrocytes, because it is well established that activation of P2X 7 causes apoptotic death of a variety of cell types (51). Recent studies have shown that an influx of Ca 2ϩ carried by Ca 2ϩ ionophores leads to phosphatidylserine externalization and the apoptosis of erythrocytes (52,53). Hypertonic shock and oxidative stress also increase phosphatidylserine externalization and erythrocyte apoptosis, and both stressors have been shown to open Ca 2ϩ -permeable channels in the membrane (53,54). These channels have many of the features of P2X 7 , because both are permeable to Ca 2ϩ and monovalent cations, and both are partly inhibited by high concentrations (1 mM) of amiloride (42,53,54). Moreover, the externalization of phosphatidylserine on erythrocytes via P2X 7 activation may lead to their subsequent phagocytosis by macrophages. Such a mechanism has been demonstrated for human erythrocytes exposed to either Ca 2ϩ ionophores or oxidative stress (52,55). In support of a role for P2X 7 in the apoptosis of erythrocytes, preliminary experiments indicate that BzATP can induce phosphatidylserine externalization on these cells. 2 Therefore, the expression of functional P2X 7 receptors on the surface of erythrocytes suggests that this receptor has a role in the apoptotic death of this cell after its 120-day life span.